Infant feeding research, particularly research looking at the impact of, and outcomes associated with breastfeeding is probably one of the most hotly debated areas of research in biological science.
It was recently claimed that a new movement of breastfeeding science denialism is gaining traction, spurred on to challenge the infallibility of breastfeeding due to a backlash against lack of support for those who wanted to breastfeed. A cynic might say that this claim is simply a thinly veiled attempt to discredit, and silence, those asking questions about the science underpinning infant feeding policy, as being motivated by revenge or self-comfort. It’s a bold statement that lumps anyone who questions infant feeding policy in with anti-vaxxers or climate change deniers.
To be a science denier is
to reject basic facts and concepts that are undisputed, well-supported parts of the scientific consensus on a subject, in favour of radical and controversial ideasWikipedia
Indeed, the main political organisations who give breastfeeding advice – UNICEF and the WHO – have made similar statements about breastfeeding and its benefits that are cited in policy documents of other organisations such as the Royal College of Paediatrics and Child Health and the Royal College of Midwives; although both do reaffirm their support for families who chose not to breastfeed. On the face of it, this may appear to be an ‘undisputed scientific consensus’ but what public health policies say and what the science says are not the same thing.
Indeed, if you dig into the data that infant feeding policy recommendations are based on, even just a wee bit, you’ll quickly discover that the scientific consensus is less than solid.
Cue Top of the Pops music for our top three reasons why the science underpinning infant feeding policy is on a bit of a shoogly peg….
In at number 3: it relies on predictions from mathematical models
Models give us a way of making predictions about what might happen if we changed a behaviour or other contributing factor. These are often used in infant feeding science to predict changes in health outcomes, health service cost savings, and even the economic impact if more women were to breastfeed for longer. However, few, if any, of these models have accounted for costs associated with breastfeeding (e.g., hospital readmissions for breastfeeding difficulties, time out of the workplace etc.) and they make assumptions that breastfeeding causes the improvements in health outcomes that the models are based on (which we’ll discover shortly is sometimes a big no-no!). The 2016 model that estimated 823,000 infant lives saved a year if we achieved ‘optimal breastfeeding rates’ has been widely cited in UK infant feeding policy. However, the study was based on low to middle income countries, and as such, is not readily applicable to a UK population.
Down one to number two: it ignores confounding variables (read the small print)
The recent WHO study on obesity is a brilliant example of this one. You see, right at the end and buried in the penultimate paragraph ‘strengths and limitations’ is a statement that the study did not consider maternal BMI and many other confounding variables that are likely to have had a big impact on the results. This of course doesn’t stop it from being used to proclaim ‘BREASTFEEDING PREVENTS OBESITY ’.
The fact is, you’ll see similar cautionary comments in many WHO documents on the impact of breastfeeding. Stating that the scientific consensus is less that solid is not as radical or controversial as some may claim . Especially when studies that take confounders into consideration suggest few differences in outcomes between breastfed and non-breastfed babies.
Top of the Pops this week at number one: it mistakes correlation for causation
This one is an oldie but goldie. The vast majority of infant feeding research is done using ‘observational studies’ where researchers observe what people do without intervening. The outcomes or behaviours of one group of people are then compared to a different group of people observed under as similar conditions as possible, to see whether there are any differences. The problem with this is that any differences seen between two groups – for example, breastfed babies and formula fed babies – may very well be because of something other than what the researchers are looking at because they’ve separated themselves into groups. This means that while researchers would love to be able to say ‘the difference is because ofbreastfeeding/formula feeding’, they can’t, because they had no control over whya family ended up feeding their babies in the way that they did. The best that anyone can say is that breastfeeding is correlated with x,y and z outcomes. Big problem here is that you can find correlations between all sorts of things.
If we want to talk about causation (because of) what we really need is a ‘randomised control trial’, such as PROBIT, where researchers are able to assign participants into groups and then measure the results. This is obviously tricky in infant feeding as it would be unethical for researchers to tell women how they should feed their babies!
The chart rundown
Science isn’t quick and it isn’t easy, infant feeding science is particularly difficult because it is so beset with potential confounding and methodological weaknesses. For many areas of infant feeding research there is no consensus and addressing this is not science denialism, indeed if it was, the WHO themselves could be accused of science denialism! In fact, failing to address these weaknesses means that the same problems that we rail against, lack of evidence-based solutions for common breastfeeding difficulties, lack of science-based solutions for comprehensive and helpful infant feeding education, will never be resolved. This is why we will always roundly reject policy that isn’t science based and enthusiastically question infant feeding science. Likewise, we will always question the science while still firmly believing that women need far better breastfeeding support. This isn’t controversial or radical.
Nobody that we know disagrees that for many families, breast will be best and they should be supported in this decision. What we reject are blanket ‘breast is best’ policies that not only hurt the families for whom breast is certainly not best, but also fail to support those for whom it is.
Policy denialism – yes, science denialism – no. This is science realism.